This is a talk given at the Redwood Center for Theoretical Neuroscience, UC Berkeley on
December 5, 2006. Speaker is Tanya Baker from the University of Chicago.

Abstract:
The dynamics of large networks of spiking neurons resembles closely that of forest fires. During such events forests contain green, burned and burning trees; likewise neural networks contain sensitive, refractory and activated neurons. We show how models of neuronal activity, both as standard forest fires as well as networks of integrate and fire neurons, are capable of displaying a variety of interesting behavior. These include long-range, power law temporal correlations in the inter-spike interval histograms, criticality, stochastic resonance in the noise driven appearance of spirals of neural activity, traveling waves, and localized clusters of activated and refractory neurons which move around. Although mean field analysis is sufficient for the characterization of population behavior away from criticality, the analysis at the critical state requires the methods of statistical neural field theory to explicitly incorporate the effects of fluctuations.